Ships shaft-rate electric field can be used as a criterion identifying the type of ship, which the mine weapons use, affect the ship safety, so the shaft-rate electric field feature extraction is very significant. In order to extract ship extremely low frequency electric field, and extract the useful frequency components in the signal by infinitely segmenting the whole band adaptively, this paper puts forward the optimal packet adaptive algorithm of harmonic wavelet. By using the algorithm to complete the shaft frequency feature extraction, the simulation results show that the proposed algorithm can effectively extract the ship shaft frequency signal, and does not require the accurate estimation of the analyzed signal frequency range, but only depends on the composition of the signal itself rather than frequency limit.